Remote Sensing Methods Outline

• Landsat-5 imagery provided by instructors (determine source from instructors)
• Use ERDAS Imagine 2010 to perform image analysis
  • Unsupervised image processing utility was used to classify image into 12 classes
  • These 12 classes were categorized one-by-one
    1. Water
    2. Roads
    3. Forest / Woods
    4. Forest / Forest
    5. Forest / Open Land
    6. Townhouse / Apartment
    7. Single Family Residential
    8. Low Density Residential
    9. Commercial
    10. Commercial / Rooftops
    11. Office / Light Industrial
    12. Agricultural
  • Google Earth was used to assist in identifying the major feature being classified
• Classified Landsat-5 image was loaded into ESRI's Arcmap 10.1 to obtain percentage of classes
  • The overall site image was clipped using the watershed boundaries for Marsh & Long Island Creek
    • 2 separate images were created; one for each watershed
  • The 12 classes were further reduced to 9 classes
    1. Water
    2. Forest
    3. Agriculture
    4. Roads
    5. Single Family Residential
    6. Low Density Residential
    7. Townhouse / Apartment
    8. Office / Light Industrial
    9. Commercial
  • Using the clipped, classified image, the total number of pixels were determined
  • The sum of each category's pixel count was determined
  • Percentages were calculated into the attribute table of the image by using the following formula:
    • PERCENTAGE = (([Classified Pixel Count] / [Total Pixel Count])*100) * [Landuse Coefficient]
  • This process was performed for each year of Landsat5 imagery.